Invited Speakers

Anna Wilbik

Maastricht University

Details

Klaus-Dieter Rest

University of Natural Resources and Life Sciences, Vienna

Details

N. N.

Institution

Details


Key Note:

Responsibility and Explainability in Automated Systems

Prof. Dr. Anna Wilbik

Professor in Data Fusion and Intelligent Interaction, Department of Advanced Computing Sciences, Faculty of Science and Engineering, Maastricht University

a.wilbik(@)maastrichtuniversity.nl

Abstract of the Talk

Recent developments in digital technology domains like Artificial Intelligence, Internet of Things and Data Science lead to possibilities for more autonomous systems that more directly interact with their business or social environment. This poses new but often implicit questions with respect to the responsibility for decisions made by these systems and the effects of these decisions on its environment. Requirements embedded in design considerations like security constraints, privacy regulations, and traceability demands, however, imply explicit questions with respect to responsibility in information processing.

Responsibility may be expected from four different classes of actors. The first perspective puts the responsibility with the system itself – thinking in terms of autonomous information systems. To support this perspective, system development methods should contribute to make responsibility an important non-functional quality attribute of the system. From a legal point of view in this perspective, the system may act as a proxy of its operational owner in decision making. The second perspective puts the responsibility with the system designer. This means that his/her use of development methods and techniques should ‘be responsible’ to arrive at trustworthy systems. The third perspective puts the responsibility with the system user, implying that we need methods that directly or indirectly support this user in using the system responsibly. The fourth perspective puts the responsibility with the party whose data is managed by the system, i.e., the object of the system, as he/she gives consent to use this system, often as part of a larger service provisioning agreement.

This talk will present the questions of the allocation of responsibility with the development and deployment of autonomous systems, based on requirements imposed by their application context. The discussion of the effect of explainability or understandability of decisions on the allocation of responsibility is an essential ingredient here. For black-box decision making-systems, the responsibility is more at the designer or system side. In the case of white-box decision-making system or a system providing explanations about its decisions, the responsibility may be shifted more towards the user and object side, as they can understand their role in the decision making process. This talk will provide the necessary concepts and models to make an analysis and design of the allocation of responsibility.

Keywords: autonomous systems, decision making, explainability, responsibility

Short CV Wilbik


Invited Talk:

Decision support for home health care services in urban regions

Mag. Dr. Klaus-Dieter Rest

Production and Logistics, University of Natural Resources and Life Sciences, Vienna

klausdieter.rest(@)boku.ac.at

Abstract of the Talk

Home health care (HHC) services allow elderly and frail people to live independently in their familiar environment and still receive professional aid. The presented work is based on my dissertation as well as on practical experience from the development of a commercial decision support tool, which has been used for several years by a major HHC service provider in Vienna. It allows to tackle the increasing demand for HHC services and to support them during daily business and in times of disasters. Vulnerability analyses were conducted to provide insights into the complex system of HHC and how it is impacted by various disasters. Causal-Loop Diagrams were used to visualize disasters (i.e., epidemics, blackouts, heatwaves, and floods) in order to identify their cascading effects and feedback loops. To support the daily planning of HHC nurses, during both daily business and in times of disasters, a new mixed integer linear program as well as several Tabu Search metaheuristics have been developed. While previous research is limited to rural areas, the focus of this work is on urban areas. The published algorithms are based on the requirements of two major HHC service providers in Vienna and cover a wide range of aspects of the daily routing of HHC nurses. It contains several working time regulations, flexible breaks, constraints regarding consistency of care, workload distribution, and synchronization. The use of time-dependent public transportation is still a unique feature. The developed decision support system is highly flexible regarding its constraints and objective function, which allows to adjust the planning according to the needs and preferences of the dispatchers. It can solve real-world sized instances within a few minutes. Comprehensive numerical studies with real-world data show significant savings, not only in travel and wait times, but also in the number of second shifts. Scenario analyses have been conducted to identify the operational limits in times of disasters.

Keywords: Home health care; Vehicle routing problem; Multimodality; Public transportation; Metaheuristic; Decision support system; Disaster management

Short CV Rest